Transcript Slide 1
Cloud Computing Imranul Hoque Today’s Cloud Computing This Talk • • • • What is cloud computing? Types of services What makes cloud computing popular? Case studies – Amazon Web Services – Eucalyptus Examples • A credit card company • A researcher running scientific computing experiments • A web-based book store • A social network startup Types of Service • Infrastructure as a Service (IaaS) – Amazon EC2, Go-Grid, Eucalyptus, etc. • Software as a Service (SaaS) – NetSuite (accounting), WorkDay (HR), etc. • Platform as a Service (PaaS) – Google AppEngine, Coghead, etc. Important Features • The illusion of infinite computing resources available on demand • The elimination of an up-front commitment by cloud users • The ability to pay for use of computing resources on a short-term basis as needed • Result: Convert capital expenditure to operational expenditure Amazon Web Services • • • • • Amazon Elastic Compute Cloud (EC2) Amazon Simple Queuing Service (SQS) Amazon Simple Storage Service (S3) Amazon Simple DB (SDB) Amazon Elastic Block Storage (EBS) Amazon Simple Storage Service • • • • • • • Programmatic access via web services API Highly scalable data storage in the cloud Simple to get going, simple to use Fast, highly available, and durable Economic: $0.15/GB-month 64 billion objects Notable users: SmugMug, Jungle Disk, Twitter, etc. Amazon EC2 • Resizable Compute Capacity – As much as you need, when you need it – Scale up or down in minutes • Complete Control via API – Create, scale, and manage instances programmatically • Variety of Instance Sizes – CPU Power, Cores, RAM, Disk • Wide Variety of Pre-built AMIs (Amazon Machine Images) – Hit the ground running with minimal system building effort • Secure and Flexible Network Security Model – Full control of access for each running instance AWS: Interesting Facts • The Washington Post – 200 EC2 instances (1,407 server hours), convert 17,481 pages of Hillary Clinton’s travel documents within 9 hours • The New York Times – 100 Amazon EC2 instances + Hadoop to convert 4TB TIFF image into 1.1 million PDFs in 24 hours • Animoto.com – 50 to 3500 servers in 3 days Eucalyptus • Open-source software infrastructure for cloud computing • Interface compatible with EC2, S3, and EBS Get me 3 small Instances 11 Architecture Cloud Controller Cluster Controller Cluster Controller Node Controller 12 Architecture (2) • Node Controller – Runs VMs – Xen/KVM – Interfaces • runInstance(), describeInstance(), terminateInstance(), describeResource(), startNetwork() Dom0 OS DomU OS DomU OS Xen Hardware 13 Architecture (3) • Cluster Controller – Access to both public and private network – Gather state information from NC – Schedule incoming VM execution request to NCs • Cloud Controller – User visible entry point – Key management – S3-interface compatible file system (Walrus) – Global scheduling 14 Event Sequence Cloud Controller Cluster Controller Node Controller Node Controller Xen Xen 15 Networking Modes • System – Use system settings • Static – Get IP from a mapping of <mac, ip> • Managed – Create a private network and get IP from that • Managed No-VLAN – Same as Managed, but no isolation 16 Internal Cluster Configuration 172.22.28.192/26 172.22.28.130 cn120 172.22.28.131 172.22.28.132 172.22.28.133 cn121 cn122 cn123 17 Internal Cluster Configuration (2) • • • • 8 Machines: 1 CC/CLC, 7 NCs OS: CentOS 5.3 (NCs running Xen kernel) Eucalyptus version: 1.5.2 Hardware: HP DL160 (2 Quad Core + 16 GB RAM + 2 TB Disk) • VM kernel: 2.6.27.21-0.1-xen • VM Image: – FC, CentOS, Ubuntu, Hadoop 18 Improvement Suggestions • Fault tolerance of cloud and cluster controller – Avoid single point of failure • Make Walrus distributed – Efficiency of access – Avoid single point of failure • Auto-scaling for Eucalyptus – Possible metrics: network, cpu, etc. 19 Demo • Admin interface – https://cn120.cloud.cs.illinois.edu:8443/ • ec2-api-tools – Version 1.3-30349 • Elasticfox – Modified for Eucalyptus • Other – Boto, Typica, Jets3t, etc. 20